You can make a full year of seasonal product photos from one good packshot: keep the product pixels, and change only the scene around it with an edit model, one season per edit. On an 8frame board the chain is the packshot, an optional cut-out, one edit node per season (Flux Edit at 7 credits for a ~1 MP input, Kontext Max at 13), and a content-preserving upscale for the winners. The limit people do not expect: Flux Edit's price rises with the size of the input image, so editing a large original costs more than editing a resized copy, and a full regeneration redraws the product as well as the scene.
TL;DR
| Step | Node or model on 8frame | Fixed price | Input |
|---|---|---|---|
| 1. Master packshot | Your photo, or Nano Banana 2 from it | 8 / 11 / 17 / 22 credits at 0.5K / 1K / 2K / 4K if generated | One sharp product photo |
| 2. Cut-out (optional) | Background removal | 12 credits per image | The packshot |
| 3. Season scene by edit | Flux Edit (FLUX.2 pro) | 7 credits for a ~1 MP input, rising with input size | Packshot plus one instruction |
| 3b. Season scene by edit, alternative | Kontext Max | 13 credits | Packshot plus one instruction |
| 3c. Season scene with props as references | Qwen Image 3 | 6 credits at 1K, 11 at 2K | 1 to 3 edit references |
| 4. Regenerate a whole scene (when editing fails) | Seedream 5 Lite | 6 credits per image | Up to 14 references |
| 5. Enlarge the winner | BRIA Increase Resolution | 6 credits, 2x or 4x | The approved scene |
Edit or regenerate: the decision that matters
There are two ways to get a product into a Christmas scene with AI, and they behave differently.
Edit keeps the image and changes part of it. You give an edit model the packshot and one instruction ("place the bottle on a snowy wooden table with warm string lights behind it, keep the bottle exactly as it is"). The model changes the surroundings; an edit is asked to leave the product pixels alone, so try it first.
Regenerate makes a new image. You give a generation model the packshot as a reference and describe the scene. The model draws everything again, including the product. You get more freedom in composition (a new angle, the product in a hand, a wider shot), but the product is now a drawing of your product, and the label is the first thing to drift.
For seasonal reskins of the same composition, edit. For a new composition, regenerate, then check the product against the real one line by line.
The board, node by node
The closest live template is PRODUCT PLACEMENT on /workflows, which places a product in a new scene. The seasonal recipe uses it once per season from the same master.
Step 1: the master packshot
Everything downstream copies this image, so it has to be the best you have: sharp at 100%, label square to the camera, one soft light, and room around the product for a scene. If the photo needs work, reshoot it before you build the board.
Decide the ratio now. If the seasonal images go to Instagram feed ads, Meta's ads guide recommends 4:5 at 1440 x 1800 pixels for feed images, and among 8frame's image generators Nano Banana is the one with a 4:5 option.
Step 2: resize a working copy
Flux Edit costs 7 credits for an input of about 1 megapixel, and the price rises with the input size (in the worst case 87 credits). A 24-megapixel camera original makes every edit expensive. Make a working copy around 1 to 2 megapixels for the edit runs, pick the winners, and only then work at full size or upscale.
Step 3: one edit node per season
Put the working copy on the left of the board and fan out one edit node per season or campaign: winter holidays, Valentine's Day, spring, summer, back to school, Black Friday. Use a board frame for each season so the drafts and the winner stay together.
Write each instruction as one change and name what must not change:
Replace the background with a summer beach scene: pale sand, soft late-afternoon sun from the left, a blurred sea in the distance. Keep the bottle, its label, its colour and its position exactly as they are. No text.
- Flux Edit (FLUX.2 pro): 7 credits for a ~1 MP input. Good first choice for background swaps.
- Kontext Max: 13 credits. A second engine when Flux Edit keeps changing the product; it can match the input image's ratio.
- Qwen Image 3: 6 credits at 1K, 11 at 2K, with 1 to 3 edit references, useful when the season needs a specific prop you have a photo of (your own gift box, a branded ribbon).
Two changes in one instruction ("make it snowy and add a gift box and change the angle") give two chances for the product to drift. If a season needs props, add them in a second edit after the background is right.
Step 4: regenerate only when editing cannot get there
Some seasonal ideas are not reskins: the product in a stocking, on a picnic blanket seen from above, held by a hand. Those need a new composition. Use Seedream 5 Lite (6 credits, up to 14 references: attach the packshot from two or three angles plus a mood reference) or Nano Banana 2. Then check the product as you would a retoucher's work: colour, label, proportions, count.
Step 5: upscale and deliver
Upscale only the approved images. BRIA Increase Resolution is 6 credits, 2x or 4x, content-preserving, up to 8192 x 8192. Topaz is 11 credits per started 24 MP block. Clarity (13 credits) is diffusion-based and always enlarges 2x on 8frame, so it can redraw a label; avoid it for products.
Seasonal ads from evergreen creative
The same board works for ads. Take your best evergreen ad image (the one already proven in your account), and make seasonal versions of it by edit, so the composition, the product and the hook stay the same and only the scene and the props change. That keeps what made the ad work and tells you whether the season itself moves the result.
Keep ad copy out of the image. Set the seasonal headline ("Gifts that ship by the 20th") in the ad platform's text fields or in a design tool. Image models redraw lettering on every generation, and a date or a discount drawn by the model is a claim you would have to check letter by letter.
If you are deciding whether to reshoot instead, what a product shoot costs is the comparison point. For a full holiday ad calendar, see the Black Friday AI ad sprint.
Which model for which step
| Need | Model | Price | What to know |
|---|---|---|---|
| Background swap, product untouched | Flux Edit | 7 for ~1 MP input, rises with size | Resize before editing |
| Background swap, second engine | Kontext Max | 13 | Can match the input's ratio; offers 4:5 |
| Season with your own prop | Qwen Image 3 | 6 at 1K, 11 at 2K | 1 to 3 edit references |
| New composition | Seedream 5 Lite | 6 | Up to 14 references |
| New composition, 4:5 feed | Nano Banana 2 | 8 / 11 / 17 / 22 by size | Has 4:5 |
| Final enlargement | BRIA | 6 | Content-preserving, 2x or 4x |
Where it breaks
- The label redraws. Even an edit can touch the product. Compare every output with the master at 100%. If the label keeps drifting, see why AI changes product labels.
- Light that disagrees. A packshot lit from the left in a scene lit from the right reads as a cut-out. Name the light direction of the original in every instruction.
- Expensive edits on originals. Flux Edit's price follows the input size. Resize first.
- Seasonal text in the image. Dates, discounts and greetings drawn by the model drift. Set them yourself.
- Too many changes per edit. One change per edit, then chain edits.
- Refusals. Almost every failure shows "Sensitive content detected. Please try changing your prompt or image." Credits for failed generations are refunded automatically; rephrase and run again.
Listing and ad rules
- Google Merchant Center: the main image must offer "an unobstructed, clean view of the item for sale", so a seasonal scene does not replace the main image. Google also has a separate attribute for lifestyle images, covered on our furniture room scenes page. AI-generated images must contain metadata saying so, and you must not remove the IPTC DigitalSourceType tag.
- TikTok ads: TikTok's Misleading and false content policy (last updated April 2026) requires an AIGC label or your own clear disclaimer for significantly edited or AI-generated content, and lists "removing or modifying backgrounds" among insignificant AI edits. A pure background swap may fall on the insignificant side; a fully generated scene with generated props is closer to AI-generated content. When in doubt, label it. More in do you have to disclose AI-generated ads.
FAQ
Can I make holiday product photos from one photo with AI?
Yes. Keep the packshot and change the scene with an edit model, one season per edit. Flux Edit is 7 credits for a ~1 MP input on 8frame, Kontext Max 13.
Should I edit or regenerate?
Edit when the composition stays the same and the product must stay pixel-close. Regenerate only for a new composition, then check the product carefully.
Why did my edit cost more than 7 credits?
Flux Edit's price rises with input size. Resize a working copy to about 1 to 2 megapixels for drafts.
Can a seasonal scene be my main product image?
Google Merchant Center asks for a clean view of the product as the main image. Use seasonal scenes as lifestyle or additional images.
Do I need to label AI background swaps in TikTok ads?
TikTok lists removing or modifying backgrounds as an insignificant AI edit, while significantly edited or fully AI-generated content needs a label or your own disclaimer. If the whole scene is generated, label it.
Sources
- Google Merchant Center Help, image requirements (clean view of the item; AI metadata and IPTC DigitalSourceType): support.google.com, checked 2026-11-09
- TikTok Ads Policy, Misleading and false content (AIGC label or own disclaimer; undisclosed AI content rejected or restricted; removing or modifying backgrounds listed as insignificant; last updated April 2026): ads.tiktok.com, checked 2026-11-09
- Meta Ads Guide, Instagram feed image ads (4:5, 1440 x 1800 recommended): facebook.com/business/ads-guide, checked 2026-11-09
- 8frame prices, defaults and limits: the 8frame app, checked 2026-11-09
Try it without signing up: the 8frame canvas opens on a real board. The canvas is free and unlimited; generation is paid from $19/month.