I made 60 product shots with Stability AI then Reels with Luma: which format got WhatsApp orders

I spent ₦15,000 on sixty product images and three weeks on twenty Reels

I sell handmade leather bags and wallets on Instagram. My customers message me on WhatsApp, I send photos or videos, they ask for the price, and if they're serious they transfer the money through their bank app. No Paystack link, no website checkout—just a chat, a bank transfer confirmation, and logistics. It's how most of us work here.

In October I had twenty products sitting in my studio with terrible phone photos. Bad lighting, cluttered background, the kind of shots that make a ₦25,000 bag look like it costs ₦8,000. I'd been putting off a proper shoot because hiring a photographer in Lagos starts at ₦50,000 for a half-day, and I didn't have the budget or the time to cart everything across town in traffic.

I decided to test two approaches. I'd generate product images with Stability AI—three angles per product, sixty shots total—and then make short Reels with Luma Ray 2 for the same twenty items. I'd post them over thirty days, track which format got more WhatsApp inquiries, and see which ones actually converted to bank transfers. Not a scientific study, just a real month of selling with two different tools and a notes app full of tally marks.

Here's what happened, and which one I'm using now.

The product shots: Stable Image Core at ₦250 per image

I used Stable Image Core through Kryotta because it handles product photography prompts better than SD 3.5 Large for this kind of work—cleaner backgrounds, more consistent lighting, less weird texture artifacts on leather. I already had decent phone photos of each bag and wallet on a white sheet. I uploaded those as reference images and wrote prompts like "professional product photography of brown leather crossbody bag on pure white background, soft studio lighting, three-quarter angle, sharp focus, commercial quality."

Sixty images cost me about ₦15,000 in total. That's roughly ₦250 per shot, and the whole batch took me four hours spread over two evenings—uploading references, tweaking prompts when the first version gave me strange stitching or unnatural shadows, exporting the final files. The results were good enough to post. Not Jumia homepage perfect, but better than my phone shots and consistent across the whole catalogue.

I posted two or three images per product across my Instagram grid and Stories, always with the price and a "DM for details" line. The images got decent engagement—likes from my existing followers, a few shares—but the WhatsApp inquiries were slower than I expected. People would save the post or send it to a friend, but they weren't messaging me immediately.

The Reels: Luma Ray 2 at three hours per video

For the Reels I used Luma Ray 2, which generates short video clips from text prompts or reference images. I gave it the same product photos and wrote prompts like "smooth rotating camera around brown leather crossbody bag on white surface, studio lighting, elegant reveal, five seconds." Each Reel took about three hours to generate, review, and edit into something I could actually post—adding text overlays with the price, a quick intro card, and exporting at the right resolution for Instagram.

Twenty Reels over three weeks. That's sixty hours of work, most of it waiting for the generation to finish and then redoing it when the camera movement looked unnatural or the bag warped halfway through. I didn't pay per Reel beyond my Kryotta workspace cost, but the time investment was real. I was generating these in the evenings after packing orders, and some nights I just didn't have the energy to babysit another render.

The Reels performed differently. They got more views—Instagram pushes video harder than stills—but the WhatsApp inquiries came faster and felt more serious. People would watch the Reel, screenshot it, and message me within an hour asking if the bag was still available. The video seemed to answer questions they'd normally ask in chat: how the strap sits, whether the leather looks thick, how the clasp works.

The numbers: response rate and actual bank transfers

I tracked every WhatsApp message that came in and whether it led to a payment. Here's the breakdown:

The Reels drove more inquiries and converted at a higher rate. But they also took fifteen times longer to produce. If I divide sales by production hours, the product shots gave me three sales per hour of work; the Reels gave me 0.5 sales per hour. That's not a perfect metric—some products just sell better—but it shows the trade-off clearly.

The product categories mattered more than I expected. Wallets and cardholders sold fine from still images. People know what a wallet looks like; they just need to see the colour and stitching quality. But bags—especially crossbody bags and totes where the shape and strap matter—converted much better from video. A five-second rotating shot showed how the bag sits and moves in a way that three static angles couldn't.

When to use stills versus video for WhatsApp commerce

If you're selling simple products where the customer already knows the form—phone cases, wallets, jewellery, notebooks—generate product shots with Stability AI and post them in batches. You'll get your catalogue online faster, and the conversion rate is good enough for items under ₦15,000 where people make quick decisions.

If you're selling anything with shape, drape, or movement—bags, clothing, shoes, anything fabric—invest the time in Reels for your best sellers. Don't do video for your whole catalogue; pick the five or six products that make up most of your revenue and make Reels for those. Use Stability AI stills for everything else.

I also learned that posting format matters on Instagram. Reels get more reach, but if someone's scrolling your grid to browse your catalogue, they want to see all the options quickly. I ended up using product shots for grid posts and Reels for Stories and the main feed when I wanted to push a specific item. The stills became my reference library; the videos became my sales tool.

One more thing: video cuts down the back-and-forth on WhatsApp. With stills, people ask "Can you send another angle?" or "Does it come with a strap?" A Reel answers most of those questions in five seconds. I saved maybe ten messages per serious inquiry, which added up when I was handling thirty chats in a weekend.

The cost breakdown: time is money, but so is your phone storage

Stability AI product shots cost me ₦15,000 for sixty images. Luma Ray 2 Reels didn't charge per video, but sixty hours of my time is worth something—if I value my labour at ₦2,000 per hour (roughly what I'd pay someone to help pack orders), that's ₦120,000 in opportunity cost. Even at half that rate, the Reels were expensive in time.

But the Reels also generated ₦775,000 in sales versus ₦300,000 from the product shots, so the return was there. The question is whether you have sixty hours to spend on twenty videos, or whether you'd rather use those hours to make more product, talk to customers, or sort out logistics.

I'm keeping both in my workflow now. I generate product shots for new items as soon as they're finished—takes me an hour to shoot references and run them through Stable Image Core, and I have posts ready for the week. Then I pick one or two hero products each month and make Reels for those. I'm not trying to make video for everything; I'm using video where it makes the biggest difference.

If you're just starting or you're trying to get a catalogue online quickly, start with Stability AI product shots. If you're already selling and you want to increase conversions on your top items, add Reels for those specific products. Don't try to do everything at once.

Questions people ask

How much does it cost to generate product images with Stability AI in Nigeria?
Using Stable Image Core through Kryotta, you're looking at roughly ₦250 per image depending on resolution and how many attempts you need. For a full product catalogue—three angles per item—budget around ₦750 per product. That's still cheaper than a photographer and faster than spending a weekend with a lightbox and your phone.

Do Instagram Reels actually convert better than product photos for WhatsApp sales?
In my test, Reels converted at 46% versus 28% for stills, but they took fifteen times longer to produce. The difference was biggest for products where shape and movement matter—bags, clothing, anything with drape. For simple items like wallets or phone cases, stills converted fine and saved me a lot of time.

Which AI video tool works best for product Reels in Nigeria?
I used Luma Ray 2 because it handles product rotation and studio lighting prompts well, and it doesn't require a high-end laptop to run. Nova Reel is faster but gave me less control over camera movement. If you're making Reels in batches, Luma's worth the extra generation time for the quality.

Should I make AI product videos for everything I sell on Instagram?
No. Make Reels for your five or six best sellers—the items that drive most of your revenue—and use Stability AI product shots for the rest of your catalogue. Video takes too long to justify for every SKU, but it makes a real difference on the products you're actively pushing.

I'm still generating product shots every week and adding Reels when I have time and a product that deserves the effort. If you're selling on Instagram and WhatsApp and you're trying to figure out where to spend your time, try both formats on a few products and track the messages. You'll know within two weeks which one works for what you sell. You can test both tools in one workspace at Kryotta—Stability AI for images and Luma Ray 2 for video, same place, no separate subscriptions.