How do you usually manage with images for an e-commerce? Does still life photography of each product cost a lot in time and money?
Well, I thought about a workflow able to generate from a basic photo of the product, a total of 5 photos from different perspectives + a 360° video.
The prompt has been planned to be as consistent as possible, but according to the possibility to start from a non-professional photo as a basis, there may be lost details. As a consequence I have imagined the need for a disclaimer on the web product page, like
*These images were generated with artificial intelligence, referring to the original product, the seller reserves the possibility of some differences in the details of the actual product
I think that this workflow may be suitable in for different product categories and different e-commerce.
Here the link to the Product e-commerce shoot workflow (https://www.morphic.com/en/workflows/019d91ac-bcb8-77f8-97ad-e345432b51b1/product-e-commerce-shoot) on Morphic
In the comments there are other examples of the outputs 😊
It’s a visual study about silence, inner freedom, and the kind of beauty that doesn’t need to shout
20
88
514
Starting with a custom subject we achieved total character consistency across every shot. By driving Kling 3.0 and Veo 3 through JSON structures, we pushed the boundaries of skin realism, maintaining hyper-realistic pores and textures.
Using technical frameworks within FLORA (https://www.linkedin.com/company/floraai/) instead of simple descriptions, we directed everything from the athlete’s speech to the physics of the AI-designed snowboard during complex tricks.
You can have a look at the workflow here: https://dub.florafauna.ai/R6Is1cB
After several (and sometimes frustrating) hours of prompting, here is the result: AI goes far beyond any limit… Yet in the end, what truly matters more and more always leads back to small, simple things — no less rich in essence for that.
Here is a minimal template, where at times I was able to use some trending AI effects from CapCut, and at other times I had to generate the results with ad hoc prompts and Seedream as the model.
I deliberately chose to use only free elements: every day I explore different kinds of systems and always try to make the most of free subscriptions, even discovering how to occasionally work around certain premium features and achieve similar results (which was also possible in this case!).
Hope you will like it 🫶
#aicapcut 🔥
Sometimes guessing camera angles with AI is a time trap. I can build the movement in 3D faster than an AI can hallucinate it. I used my own 3D wireframe as a 'digital skeleton' to keep control over timing and motion, leaving the AI to focus on adapting different styles to the concept.