Experimenting with building a consistent photographic style across different subjects and angles, not one character, but one shoot.
Goal: same light, same mood, every frame, regardless of who's in it. Used one style reference image + prompt variations in @floraai to get this set.
Took 3 reference photos and swapped in the character I built in https://t.co/aChTBfJ9ch kept the scene, the light, the composition. Just replaced the person.
Goal: see if the swap holds up across completely different lighting.
Experimenting with creating a realistic character and placing it in the same scene across different poses.
Goal: realism and a natural feel. Used the character lock technique in @floraai + reference image to get this first set.
we’re in this era of personal software and figuring out how to make outputs more deterministic. but with more creative things like image generation, I find it's finding the sweet spot between control and surprise.
so crazy how now you start seeing everything as context and input; now whenever i find a good reference, i’m just itching to plug it in and see what it spits back.
(made in between waiting for codex and claude, visual style ref sourced from a book, cupid in lower right from running eros, holding a torch statue spotted irl at the jp morgan library)
@JordanGilroy CRO is a pretty common use case.
I work with Shopify brands and often design experiment variations based on the live site. We rarely have access to the original Figma files, so tools like this help us iterate on the current experience without rebuilding everything from scratch.
@figma I gave it a try and, so far:
• No image export
• Original site fonts aren’t preserved
• No option to select different viewport sizes
Would love to see those things improved. For now, I’m sticking with https://t.co/PkTd3tIWAe by @divriots.